No spooky action at a distance

[Joy Christian]

We now set the polarizer PB offset by 90 ° to PA on alpha + 90 °.

You are not allowed to do that. That "offset" can be done only nonlocally. There is no way for Bob to know how to offset PB by 90 ° without knowing what angle Alice has chosen for PA.

***

[Esail]

We now set the polarizer PB offset by 90 ° to PA on alpha + 90 °.

You are not allowed to do that. That "offset" can be done only nonlocally. There is no way for Bob to know how to offset PB by 90 ° without knowing what angle Alice has chosen for PA.

***

Bob is free to choose any setting of his polarizer. Only in case, he did chose alpha+90° the experimenter afterwards would detect 100% coincidence between Alice's and Bob's findings. There is nothing nonlocal.
But what we can say is that a selection of photons with particular lambda values done by Alice's polarizer does also mean of a selection of Bob's photons with the same lambda values as the A and B photons share the same value of lambda.

[Joy Christian]

We now set the polarizer PB offset by 90 ° to PA on alpha + 90 °.

You are not allowed to do that. That "offset" can be done only nonlocally. There is no way for Bob to know how to offset PB by 90 ° without knowing what angle Alice has chosen for PA.

Bob is free to choose any setting of his polarizer. Only in case, he did chose alpha+90° the experimenter afterwards would detect 100% coincidence between Alice's and Bob's findings. There is nothing nonlocal.
But what we can say is that a selection of photons with particular lambda values done by Alice's polarizer does also mean of a selection of Bob's photons with the same lambda values as the A and B photons share the same value of lambda.

If there is nothing nonlocal in any of the prescriptions you have described, then you should be able to write down two functions A(a, lambda) and B(b, lambda) with completely independent settings "a" and "b". Something in your model seems to prevent you from doing that. That can only mean that there is some kind of nonlocality hidden in your complicated prescriptions.

***

[FrediFizzx]

...
You are not allowed to do that. That "offset" can be done only nonlocally. There is no way for Bob to know how to offset PB by 90 ° without knowing what angle Alice has chosen for PA.

Bob is free to choose any setting of his polarizer. Only in case, he did chose alpha+90° the experimenter afterwards would detect 100% coincidence between Alice's and Bob's findings. There is nothing nonlocal.
But what we can say is that a selection of photons with particular lambda values done by Alice's polarizer does also mean of a selection of Bob's photons with the same lambda values as the A and B photons share the same value of lambda.

If there is nothing nonlocal in any of the prescriptions you have described, then you should be able to write down two functions A(a, lambda) and B(b, lambda) with completely independent settings "a" and "b". Something in your model seems to prevent you from doing that. That can only mean that there is some kind of nonlocality hidden in your complicated prescriptions.

***

Yeah, I can't make heads or tails of what he is doing either. He has A(delta, lambda) and B(delta, lambda). Clearly non-local if the delta's are the same. Only lambda can be the same.
.

[Esail]

If there is nothing nonlocal in any of the prescriptions you have described, then you should be able to write down two functions A(a, lambda) and B(b, lambda) with completely independent settings "a" and "b". Something in your model seems to prevent you from doing that. That can only mean that there is some kind of nonlocality hidden in your complicated prescriptions.

***

The function A(alpha,lambda) is defined in my recent post. A complete evaluation is given in the program spec above.

[Esail]

Yeah, I can't make heads or tails of what he is doing either. He has A(delta, lambda) and B(delta, lambda). Clearly non-local if the delta's are the same. Only lambda can be the same.
.

Why can't delta be the same? Delta is the local angle between the local polarizer setting and the local polarization of the respective photon. If the A-photon has the polarization 0° and the PA setting is alpha then delta at A is alpha-0° = alpha. If the A-photon has polarization 0° than the B-photon has polarization 90° from the initial conditions. Setting polarizer PB to alpha+ 90° we get delta at B = alpha + 90°- 90° = alpha as well. What is curious about this?

[gill1109]

Yeah, I can't make heads or tails of what he is doing either. He has A(delta, lambda) and B(delta, lambda). Clearly non-local if the delta's are the same. Only lambda can be the same.
.

Why can't delta be the same? Delta is the local angle between the local polarizer setting and the local polarization of the respective photon. If the A-photon has the polarization 0° and the PA setting is alpha then delta at A is alpha-0° = alpha. If the A-photon has polarization 0° than the B-photon has polarization 90° from the initial conditions. Setting polarizer PB to alpha+ 90° we get delta at B = alpha + 90°- 90° = alpha as well. What is curious about this?

If both settings are known in advance, nothing non-local need occur. Any correlations can be locally made to exist between Alice and Bob's observations. Your problem is how to generate the singlet correlations when Alice's setting is created by some random (unpredictable) process so late in time, that there is no way that its value could be "available" to the physical stuff over at Bob's place at the moment when his measurement is completed and a macroscopic outcome is fixed (macroscopic, e.g., engraved on a tablet of stone). Your program sketch fails to win my challenge.

[Esail]

If both settings are known in advance, nothing non-local need occur. Any correlations can be locally made to exist between Alice and Bob's observations. Your problem is how to generate the singlet correlations when Alice's setting is created by some random (unpredictable) process so late in time, that there is no way that its value could be "available" to the physical stuff over at Bob's place at the moment when his measurement is completed and a macroscopic outcome is fixed (macroscopic, e.g., engraved on a tablet of stone). Your program sketch fails to win my challenge.

Unfortunately for you, the measurement results only depend on the last polarizer setting, which ultimately makes the selection. This is completely clear because the measurement results are predetermined. What happened before is irrelevant. This is the case in QM - see the work of G. Weihs - and also in my model. So give me the money.

[gill1109]

If both settings are known in advance, nothing non-local need occur. Any correlations can be locally made to exist between Alice and Bob's observations. Your problem is how to generate the singlet correlations when Alice's setting is created by some random (unpredictable) process so late in time, that there is no way that its value could be "available" to the physical stuff over at Bob's place at the moment when his measurement is completed and a macroscopic outcome is fixed (macroscopic, e.g., engraved on a tablet of stone). Your program sketch fails to win my challenge.

Unfortunately for you, the measurement results only depend on the last polarizer setting, which ultimately makes the selection. This is completely clear because the measurement results are predetermined. What happened before is irrelevant. This is the case in QM - see the work of G. Weihs - and also in my model. So give me the money.

The deal I am offering you is that you must program your method in a computer language of your choice but preferably one which many people can easily run on their own computer (Apple, Windows, or Linux). I may supply arbitrary lists of settings for Alice and Bob (both binary) of any length. Your program must then output lists of binary outcomes of the same length. If your program uses a random number generator then it must include facilities to set the seed and to save the seed. What is your objection to this? A real experimenter can choose just how many trials they are going to do. A real experimenter doing a standard Bell-type experiment chooses settings using a random number generator. In computer simulation programs it is essential to be able to run the program any number of times with the same stream of pseudo random numbers, or pre-stored physical random numbers, for debugging purposes.

I need these facilities so that I can check, by some random checks, that the outcome of (e.g.) Alice's 100th measurement does not depend on Bob's 100th setting, given what happened in the first 99 trials. If you can satisfy these constraints *and* violate Bell-CHSH inequality systematically then I will send you a check, and I will also tell the world that you succeeded, where till now nobody has succeeded. The amount of money I can give you depends on the size of the sample and the size of the deviation. I need to be fairly confident that I am not going to lose. You need not pay me anything, you simply either win or lose. Everyone in the world will be able to see whether you won or lost, since we will post the results on this (or another suitable) forum.

[gill1109]

Esai's program is completely deterministic. You must start by deciding whether you are doing the spin 1/2 or the spin 1 case. You also fix the two settings. You fix n = 1000. Then you do twice 1 + n = 1001 measurements on one particle. The outcomes are deterministic. The "hidden variable" simply takes values 0/n, 1/n, ..., n/n. This is done twice. One polarizer is given the original orientation assigned to that polarizer, the other is given the difference between the two orientations. It is a very naive simulation of the known quantum mechanical probability distribution - two settings, two measurements, two binary outcomes, four probabilities. Esai knows those probabilities and puts them into his program.

[FrediFizzx]

If both settings are known in advance, nothing non-local need occur. Any correlations can be locally made to exist between Alice and Bob's observations. Your problem is how to generate the singlet correlations when Alice's setting is created by some random (unpredictable) process so late in time, that there is no way that its value could be "available" to the physical stuff over at Bob's place at the moment when his measurement is completed and a macroscopic outcome is fixed (macroscopic, e.g., engraved on a tablet of stone). Your program sketch fails to win my challenge.

Unfortunately for you, the measurement results only depend on the last polarizer setting, which ultimately makes the selection. This is completely clear because the measurement results are predetermined. What happened before is irrelevant. This is the case in QM - see the work of G. Weihs - and also in my model. So give me the money.

The deal I am offering you is that you must program your method in a computer language of your choice but preferably one which many people can easily run on their own computer (Apple, Windows, or Linux). I may supply arbitrary lists of settings for Alice and Bob (both binary) of any length. Your program must then output lists of binary outcomes of the same length. If your program uses a random number generator then it must include facilities to set the seed and to save the seed. What is your objection to this? A real experimenter can choose just how many trials they are going to do. A real experimenter doing a standard Bell-type experiment chooses settings using a random number generator. In computer simulation programs it is essential to be able to run the program any number of times with the same stream of pseudo random numbers, or pre-stored physical random numbers, for debugging purposes.

I need these facilities so that I can check, by some random checks, that the outcome of (e.g.) Alice's 100th measurement does not depend on Bob's 100th setting, given what happened in the first 99 trials. If you can satisfy these constraints *and* violate Bell-CHSH inequality systematically then I will send you a check, and I will also tell the world that you succeeded, where till now nobody has succeeded. The amount of money I can give you depends on the size of the sample and the size of the deviation. I need to be fairly confident that I am not going to lose. You need not pay me anything, you simply either win or lose. Everyone in the world will be able to see whether you won or lost, since we will post the results on this (or another suitable) forum.

You are still mixing up Bell's junk physics theory with Gill's "theorem".
.

[FrediFizzx]

...
The deal I am offering you is that you must program your method in a computer language of your choice but preferably one which many people can easily run on their own computer (Apple, Windows, or Linux). I may supply arbitrary lists of settings for Alice and Bob (both binary) of any length. Your program must then output lists of binary outcomes of the same length. If your program uses a random number generator then it must include facilities to set the seed and to save the seed. What is your objection to this? A real experimenter can choose just how many trials they are going to do. A real experimenter doing a standard Bell-type experiment chooses settings using a random number generator. In computer simulation programs it is essential to be able to run the program any number of times with the same stream of pseudo random numbers, or pre-stored physical random numbers, for debugging purposes.

I need these facilities so that I can check, by some random checks, that the outcome of (e.g.) Alice's 100th measurement does not depend on Bob's 100th setting, given what happened in the first 99 trials. If you can satisfy these constraints *and* violate Bell-CHSH inequality systematically then I will send you a check, and I will also tell the world that you succeeded, where till now nobody has succeeded. The amount of money I can give you depends on the size of the sample and the size of the deviation. I need to be fairly confident that I am not going to lose. You need not pay me anything, you simply either win or lose. Everyone in the world will be able to see whether you won or lost, since we will post the results on this (or another suitable) forum.

You are still mixing up Bell's junk physics theory with Gill's "theorem".
.

So, what exactly is Gill's "theorem"? I suppose it would go something like this in words; "No Local-Realistic hidden variable theory can simulate an EPR type experiment." But quantum mechanics can't do that either so what is the point? Plus Jay Yablon has successfully demonstrated that QM is local for the EPR-Bohm scenario. Makes sense; all action of this type in Nature has to be local. The the "local" part is a consideration since if you make local measurement functions for QM, you won't be able to simulate an EPR type experiment either.
.

[Esail]

Esai's program is completely deterministic. You must start by deciding whether you are doing the spin 1/2 or the spin 1 case. You also fix the two settings. You fix n = 1000. Then you do twice 1 + n = 1001 measurements on one particle. The outcomes are deterministic. The "hidden variable" simply takes values 0/n, 1/n, ..., n/n. This is done twice. One polarizer is given the original orientation assigned to that polarizer, the other is given the difference between the two orientations. It is a very naive simulation of the known quantum mechanical probability distribution - two settings, two measurements, two binary outcomes, four probabilities. Esai knows those probabilities and puts them into his program.

Implemented in the program is the model derived in my paper. The measurement values for A and B are first determined using the initial state eg. photonpairs A/B with 0°/90° and 90°/0° polarization in equal shares. Subsequently it is calculated for side B if the detected photon at beta would hit a polarizer on side B which is set perpendicular to alpha at side A. What is wrong with this?

[gill1109]

Esai's program is completely deterministic. You must start by deciding whether you are doing the spin 1/2 or the spin 1 case. You also fix the two settings. You fix n = 1000. Then you do twice 1 + n = 1001 measurements on one particle. The outcomes are deterministic. The "hidden variable" simply takes values 0/n, 1/n, ..., n/n. This is done twice. One polarizer is given the original orientation assigned to that polarizer, the other is given the difference between the two orientations. It is a very naive simulation of the known quantum mechanical probability distribution - two settings, two measurements, two binary outcomes, four probabilities. Esai knows those probabilities and puts them into his program.

Implemented in the program is the model derived in my paper. The measurement values for A and B are first determined using the initial state eg. photonpairs A/B with 0°/90° and 90°/0° polarization in equal shares. Subsequently it is calculated for side B if the detected photon at beta would hit a polarizer on side B which is set perpendicular to alpha at side A. What is wrong with this?

The problem is that your model cannot be implemented on a network of computes whose communications are constrained so as to reflect the spatial and temporal separations of a real Bell-type experiment. Alice and Bob receive photons from a source. Alice and Bob receive settings from "outside". Alice and Bob see outcomes. Alice sees her outcome before Bob's setting could possibly have been transmitted in any way from Bob's side of the experiment to Alice's.

Bell's theorem is a simple mathematical theorem which says that it can't be done. Of course, there may be something wrong in the proof of that theorem. I don't think so, for one moment. But in principle, it is possible, that the entire community of physicists and mathematicians have overlooked something for 50 years. If so, it should be easy for you to deliver the goods. Give us the computer programs and let us test them.

Implement your "program" in a real program which I can run in e.g. Python or R, and let me test it. I've told you what kind of tests I will do. I've looked at your "pseudo code" (seems to be some kind of Basic, that's not a good sign, it is a computer language of the 60s which does not exactly encourage good programming practices!). I get the idea you don't understand what you are up against. Doesn't matter.

[Esail]

The problem is that your model cannot be implemented on a network of computes whose communications are constrained so as to reflect the spatial and temporal separations of a real Bell-type experiment. Alice and Bob receive photons from a source. Alice and Bob receive settings from "outside". Alice and Bob see outcomes. Alice sees her outcome before Bob's setting could possibly have been transmitted in any way from Bob's side of the experiment to Alice's.

Bell's theorem is a simple mathematical theorem which says that it can't be done. Of course, there may be something wrong in the proof of that theorem. I don't think so, for one moment. But in principle, it is possible, that the entire community of physicists and mathematicians have overlooked something for 50 years. If so, it should be easy for you to deliver the goods. Give us the computer programs and let us test them.

Implement your "program" in a real program which I can run in e.g. Python or R, and let me test it. I've told you what kind of tests I will do. I've looked at your "pseudo code" (seems to be some kind of Basic, that's not a good sign, it is a computer language of the 60s which does not exactly encourage good programming practices!). I get the idea you don't understand what you are up against. Doesn't matter.

My intention is to dicuss the the clues of the theory rather than computer programs. However, I've written a spec which consists of three parts, one for the source and the other two for Wing A and B repectively. The source produces randomly photons with polarization 0° and 90° and random lambda which are sent to the wings. The output of the wings ist stored together with an order number in order to identify matches.
Here follows the spec:

HV model for the singlet state

Program for Spin1:
1. Theory
• Boundary conditions: photon pairs 0 ° / 90 ° and 90 ° / 0 ° in equal shares.
• Photons A and B of a pair have the same property Lambda
• Set polarizer PA to alpha and polarizer PB to beta
• (alpha and beta between 0 and 180°)
• Polarizer A selects A-photons with p-state alpha
• This selection means a selection of the associated B-photons in p-state and polarization alpha+pi/2
• Polarizer B selects B-photons with p-state and polarization beta.

SEPP Source:
2. For i=1,n
3. Lambda = random1(0,1)
4. Polarization= 0
5. If random2(0,1) >0,5 then Polarization=90
6. Send (lambda, Polarization)

Wing A
7. Initialize only once CountA := 0

8. Read (lambda, Polarization)
9. If Polarization = 0 then ind=1 else ind=-1
10. deltaA := alpha -Polarization
11. Call Hit(ResultA,deltaA,lambda,countA,ind)
12. If ResultA=1 then
13. Store (ResultA, countA)

Wing B
14. Initialize only once CountA := countB := 0
15. deltaB := beta – alpha- pi/2

16. Read (lambda, Polarization)
17. If Polarization = 0 then ind=1 else ind=-1
18. deltaA := alpha -Polarization
19. Call Hit(ResultA,deltaA,lambda,countA,ind)
20. If ResultA=1 then
21. Call Hit(ResultB,deltaB,lambda,countB,ind)
22. Store (ResultA, ResultB, countA)
23. Conditional Probability P = CountB / CountA




a. Subroutine: Hit(Result,delta,lambda,count,ind)
b. Result=-1
c. If delta <0 then delta = delta+pi
d. If ind=1 then
e. If 0<delta<pi/2 or pi<delta<3/2pi then
f. If lambda <cos2 (delta) Then set Result := 1 and count := count + 1
g. Exit
h. Else
i. If lambda >sin2 (delta) Then set Result := 1 and count := count + 1
j. Exit

k. If ind= -1 then
l. If 0<delta<pi/2 or pi<delta<3/2pi then
m. If lambda > cos2 (delta) Then set Result := 1 and count := count + 1
n. Exit
o. Else
p. If lambda <sin2 (delta) Then set Result := 1 and count := count + 1
q. Exit
r. end

[Heinera]

[...]
15. deltaB := beta – alpha- pi/2
[...]

Wing B should know nothing about alpha. This is non-local.

[Joy Christian]

[...]
15. deltaB := beta – alpha- pi/2
[...]

Wing B should know nothing about alpha. This is non-local.

I have been saying this for ages, but he does not accept that for some reason.

***

[Esail]

[...]
15. deltaB := beta – alpha- pi/2
[...]

Wing B should know nothing about alpha. This is non-local.

I have been saying this for ages, but he does not accept that for some reason.

***

If photons at wing A are selected with alpha the thus selected peer photons at wing B are in state alpha+pi/2 as described in the paper.

[FrediFizzx]

[...]
15. deltaB := beta – alpha- pi/2
[...]

Wing B should know nothing about alpha. This is non-local.

I have been saying this for ages, but he does not accept that for some reason.

***

If photons at wing A are selected with alpha the thus selected peer photons at wing B are in state alpha+pi/2 as described in the paper.

Doesn't matter. It is still non-local. Try again.
.

[Heinera]

If photons at wing A are selected with alpha the thus selected peer photons at wing B are in state alpha+pi/2 as described in the paper.

Locality means that the result of measurement at wing B should not in any way depend on the setting alpha. Anything else is non-local.